jnthn RPA wasn't sparse by default. The big thing QRPA manages better is shift, which it turns out is important in the current list model (not to mention various common idioms folks might write) 10:29
timotimo is the sparseness thing very problematic? 11:25
[Coke] not unless we want to support sparse arrays, I imagine. 12:24
hoelzro FROGGS: I tried debugging the ModuleLoader a bit (regarding that lazy deserialization issue I found), but that didn't reveal much 14:30
FROGGS :/
that's sad
hoelzro and I don't understand MoarVM well enough to be able to make sense of jnthn's commit =/
jnthn hoelzro: I understand my commit, kinda...but I struggle a little more on why it changed things the way you're seeing. 14:31
hoelzro I'm wondering if it would be ok if we added an env var that turned off lazy deserialization for now
=(
then we're doomed!
DOOOOOOMED 14:32
hoelzro while we're doomed, anyone mind if I add such a flag? 14:32
(assuming I can figure out how)
tadzik FROGGS: come back, it's not that bad!
FROGGS[mobile] *g* 14:34
jnthn hoelzro: What is odd about it is that if anything is repossessed, we already should not be trying to lazily deserialize those things 14:36
hoelzro: As in, I put in code that should be handling the case that appears problematic...
hoelzro eesh 14:37
it's almost like when the second module is loaded, the WHO for Example is being replaced by a clean slate 14:38
but I've checked, and that remains the same
but I'm wondering if some structure deeper down that does the actual containing within the WHO is being replaced 14:39
jnthn If you put in such a flag, then one thing to try is what resolve_repossessions gets with and without it in the Perl 6 module loader
hoelzro alright 14:43
I guess the silver lining here is I get to learn how module loading works =P
jnthn Thankfully, module loading is mostly quite boring compared to the serializer :) 14:44
There were some notes on it in the internals course, iirc
hoelzro I should review those, then 14:45
nwc10 jnthn: flapping ASAN barfage for t/spec/S17-procasync/print.rakudo.moar -- paste.scsys.co.uk/425134 19:38
not sure when this started, or how to golf it
lizmat nwc10: maybe as soon as I wrote the test? which is not too long ago? 19:40
nwc10 lizmat: that would be a reasonable answer. :-/
I'm afraid that I don't have anything useful to add, and that I'm about to disappear off to bed 19:41
lizmat well, I think the barfage is correct, as that test flaps as hell
so I wouldn't be surprised if the ASAN flapping is in sync
good night, nwc10! 19:42